Zheng Zhiming

Jingcheng Zhou, Wei Wei, Xing Li et al.

Deep learning utilizing deep neural networks (DNNs) has achieved a lot of success recently in many important areas such as computer vision, natural language processing, and recommendation systems. The lack of convexity for DNNs has been seen as a major disadvantage of many optimization methods, such as stochastic gradient descent, which greatly reduces the genelization of neural network applications. We realize that the convexity make sense in the neural network and propose the exponential multilayer neural network (EMLP), a class of parameter convex neural network (PCNN) which is convex with regard to the parameters of the neural network under some conditions that can be realized. Besides, we propose the convexity metric for the two-layer EGCN and test the accuracy when the convexity metric changes. For late experiments, we use the same architecture to make the exponential graph convolutional network (EGCN) and do the experiment on the graph classificaion dataset in which our model EGCN performs better than the graph convolutional network (GCN) and the graph attention network (GAT).

Miao Shuyi, Qiu Wangjie, Tu Xiaofan et al.

With the rapid advancement of Web 3.0 technologies, public blockchain platforms are witnessing the emergence of novel services designed to enhance user privacy and anonymity. However, the powerful untraceability features inherent in these services inadvertently make them attractive tools for criminals seeking to launder illicit funds. Notably, existing de-anonymization methods face three major challenges when dealing with such transactions: highly homogenized transactional semantics, limited ability to model temporal discontinuities, and insufficient consideration of structural sparsity in account association graphs. To address these, we propose GradWATCH, designed to track anonymous accounts in Ethereum privacy-preserving services. Specifically, we first design a learnable account feature mapping module to extract informative transactional semantics from raw on-chain data. We then incorporate transaction relations into the account association graph to alleviate the adverse effects of structural sparsity. To capture temporal evolution, we further propose an edge-aware sliding-window mechanism that propagates and updates gradients at three granularities. Finally, we identify accounts controlled by the same entity by measuring their embedding distances in the learned representation space. Experimental results show that even under the conditions of unbalanced labels and sparse transactions, GradWATCH still achieves significant performance gains, with relative improvements ranging from 1.62% to 15. 22% in the MRR and from 3. 85% to 7. 31% in the F_1.

Jinwen Chen, Hainan Zhang, Liang Pang et al.

The current RAG system requires uploading plaintext documents to the cloud, risking private data leakage. Parametric RAG (PRAG) addresses this by encoding documents as LoRA within LLMs, enabling reasoning without exposing raw content. However, it still faces two issues: (1) PRAG demands synthesizing QA pairs and fine-tuning LLM for each individual document to create its corresponding LoRA, leading to unacceptable inference latency. (2) The performance of PRAG relies solely on synthetic QA data, lacking internal alignment with standard RAG, resulting in poor generalization on out-of-distribution(OOD) inputs. Therefore, achieving high-efficiency parameterization while maintaining RAG-level performance remains a critical challenge for privacy-preserving reasoning. In this paper, we propose DistilledPRAG, a generalizable knowledge-distilled parametric RAG model aligned with standard RAG in document structure and parameter activation. We first synthesize QA pairs from single and multi-documents to enhance cross-document reasoning. Then, we mask the plaintext documents with a special token and translate them to LoRA via a parameter generator, maintaining the standard RAG document structure. Finally, guided by synthetic QA data, we train the parameter generator to match standard RAG's hidden states and output logits, enabling RAG-style reasoning without original documents. Experiments on four QA datasets show that DistilledPRAG outperforms baselines in accuracy and generalizes well on OOD data.